Upload folder using huggingface_hub

asr_diarization/pipeline.py

@@ -5,6 +5,7 @@ import torchaudio
 import noisereduce as nr
 from pyannote.audio import Pipeline
 from transformers import pipeline as hf_pipeline
+from transformers import WhisperForConditionalGeneration, WhisperProcessor
 import tempfile
 
 from pyannote.core import Annotation, Segment
@@ -25,7 +26,7 @@ class ASR_Diarization:
         # Load ASR model with timestamps
         self.asr_pipeline = hf_pipeline(
             "automatic-speech-recognition",
-            model=asr_model,
+            model=WhisperForConditionalGeneration.from_pretrained(asr_model),
             device=0 if self.device == "cuda" else -1,
             return_timestamps=True
         )